Protein-ligand binding affinity prediction using multi-instance learning with docking structures

IntroductionRecent advances in 3D structure-based deep learning approaches demonstrate improved accuracy in predicting protein-ligand binding affinity in drug discovery. These methods complement physics-based computational modeling such as molecular docking for virtual high-throughput screening. Des...

Full description

Saved in:
Bibliographic Details
Main Authors: Hyojin Kim, Heesung Shim, Aditya Ranganath, Stewart He, Garrett Stevenson, Jonathan E. Allen
Format: Article
Language:English
Published: Frontiers Media S.A. 2025-01-01
Series:Frontiers in Pharmacology
Subjects:
Online Access:https://www.frontiersin.org/articles/10.3389/fphar.2024.1518875/full
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850087480980668416
author Hyojin Kim
Heesung Shim
Aditya Ranganath
Stewart He
Garrett Stevenson
Jonathan E. Allen
author_facet Hyojin Kim
Heesung Shim
Aditya Ranganath
Stewart He
Garrett Stevenson
Jonathan E. Allen
author_sort Hyojin Kim
collection DOAJ
description IntroductionRecent advances in 3D structure-based deep learning approaches demonstrate improved accuracy in predicting protein-ligand binding affinity in drug discovery. These methods complement physics-based computational modeling such as molecular docking for virtual high-throughput screening. Despite recent advances and improved predictive performance, most methods in this category primarily rely on utilizing co-crystal complex structures and experimentally measured binding affinities as both input and output data for model training. Nevertheless, co-crystal complex structures are not readily available and the inaccurate predicted structures from molecular docking can degrade the accuracy of the machine learning methods.MethodsWe introduce a novel structure-based inference method utilizing multiple molecular docking poses for each complex entity. Our proposed method employs multi-instance learning with an attention network to predict binding affinity from a collection of docking poses.ResultsWe validate our method using multiple datasets, including PDBbind and compounds targeting the main protease of SARS-CoV-2. The results demonstrate that our method leveraging docking poses is competitive with other state-of-the-art inference models that depend on co-crystal structures.DiscussionThis method offers binding affinity prediction without requiring co-crystal structures, thereby increasing its applicability to protein targets lacking such data.
format Article
id doaj-art-8b6ab930ce45489980464c1ea6a967ed
institution DOAJ
issn 1663-9812
language English
publishDate 2025-01-01
publisher Frontiers Media S.A.
record_format Article
series Frontiers in Pharmacology
spelling doaj-art-8b6ab930ce45489980464c1ea6a967ed2025-08-20T02:43:12ZengFrontiers Media S.A.Frontiers in Pharmacology1663-98122025-01-011510.3389/fphar.2024.15188751518875Protein-ligand binding affinity prediction using multi-instance learning with docking structuresHyojin Kim0Heesung Shim1Aditya Ranganath2Stewart He3Garrett Stevenson4Jonathan E. Allen5Center for Applied Scientific Computing, Lawrence Livermore National Laboratory, Livermore, CA, United StatesBiosciences and Biotechnology Division, Lawrence Livermore National Laboratory, Livermore, CA, United StatesCenter for Applied Scientific Computing, Lawrence Livermore National Laboratory, Livermore, CA, United StatesGlobal Security Computing Applications Division, Lawrence Livermore National Laboratory, Livermore, CA, United StatesComputational Engineering Division, Lawrence Livermore National Laboratory, Livermore, CA, United StatesGlobal Security Computing Applications Division, Lawrence Livermore National Laboratory, Livermore, CA, United StatesIntroductionRecent advances in 3D structure-based deep learning approaches demonstrate improved accuracy in predicting protein-ligand binding affinity in drug discovery. These methods complement physics-based computational modeling such as molecular docking for virtual high-throughput screening. Despite recent advances and improved predictive performance, most methods in this category primarily rely on utilizing co-crystal complex structures and experimentally measured binding affinities as both input and output data for model training. Nevertheless, co-crystal complex structures are not readily available and the inaccurate predicted structures from molecular docking can degrade the accuracy of the machine learning methods.MethodsWe introduce a novel structure-based inference method utilizing multiple molecular docking poses for each complex entity. Our proposed method employs multi-instance learning with an attention network to predict binding affinity from a collection of docking poses.ResultsWe validate our method using multiple datasets, including PDBbind and compounds targeting the main protease of SARS-CoV-2. The results demonstrate that our method leveraging docking poses is competitive with other state-of-the-art inference models that depend on co-crystal structures.DiscussionThis method offers binding affinity prediction without requiring co-crystal structures, thereby increasing its applicability to protein targets lacking such data.https://www.frontiersin.org/articles/10.3389/fphar.2024.1518875/fullAI-driven drug developmentvirtual high-throughput screeningprotein-ligand interactionmolecular docking3D atomic graph representationstructure-based machine learning
spellingShingle Hyojin Kim
Heesung Shim
Aditya Ranganath
Stewart He
Garrett Stevenson
Jonathan E. Allen
Protein-ligand binding affinity prediction using multi-instance learning with docking structures
Frontiers in Pharmacology
AI-driven drug development
virtual high-throughput screening
protein-ligand interaction
molecular docking
3D atomic graph representation
structure-based machine learning
title Protein-ligand binding affinity prediction using multi-instance learning with docking structures
title_full Protein-ligand binding affinity prediction using multi-instance learning with docking structures
title_fullStr Protein-ligand binding affinity prediction using multi-instance learning with docking structures
title_full_unstemmed Protein-ligand binding affinity prediction using multi-instance learning with docking structures
title_short Protein-ligand binding affinity prediction using multi-instance learning with docking structures
title_sort protein ligand binding affinity prediction using multi instance learning with docking structures
topic AI-driven drug development
virtual high-throughput screening
protein-ligand interaction
molecular docking
3D atomic graph representation
structure-based machine learning
url https://www.frontiersin.org/articles/10.3389/fphar.2024.1518875/full
work_keys_str_mv AT hyojinkim proteinligandbindingaffinitypredictionusingmultiinstancelearningwithdockingstructures
AT heesungshim proteinligandbindingaffinitypredictionusingmultiinstancelearningwithdockingstructures
AT adityaranganath proteinligandbindingaffinitypredictionusingmultiinstancelearningwithdockingstructures
AT stewarthe proteinligandbindingaffinitypredictionusingmultiinstancelearningwithdockingstructures
AT garrettstevenson proteinligandbindingaffinitypredictionusingmultiinstancelearningwithdockingstructures
AT jonathaneallen proteinligandbindingaffinitypredictionusingmultiinstancelearningwithdockingstructures